Macrophages employ quorum licensing to regulate collective activation

Abstract

Macrophage-initiated inflammation is tightly regulated to eliminate threats such as infections while suppressing harmful immune activation. However, individual cells' signaling responses to pro-inflammatory cues are heterogeneous, with subpopulations emerging with high or low activation states. Here, we use single-cell tracking and dynamical modeling to develop and validate a revised model for lipopolysaccharide (LPS)-induced macrophage activation that invokes a mechanism we term quorum licensing. The results show that bimodal phenotypic partitioning of macrophages is primed during the resting state, dependent on cumulative history of cell density, predicted by extrinsic noise in transcription factor expression, and independent of canonical LPS-induced intercellular feedback in the tumor necrosis factor (TNF) response. Our analysis shows how this density-dependent coupling produces a nonlinear effect on collective TNF production. We speculate that by linking macrophage density to activation, this mechanism could amplify local responses to threats and prevent false alarms.

Fig. 1. The TNF response to LPS is heterogeneous and requires intercellular communication.

a The diagram summarizes the perturbations and stimuli applied to investigate TNF expression and intercellular communication (hps, hours post-stimulation with LPS). LPS activates TLR4 signaling, which induces TNF expression. IL-10 pretreatment activates IL-10R signaling, which inhibits LPS-induced TNF expression. Secreted TNF activates TNFR signaling, which induces TNF further through intercellular feedback. BFA prevents secretion, causing TNF to accumulate intracellularly. Varying the cell density modulates the concentrations of secreted factors such as TNF. sTNFR binds extracellular TNF and prevents TNFR signaling. b IL-10 pretreatment diminishes LPS-induced TNF expression. c TNF expression is heterogeneous with high-expressing and low-expressing subpopulations. After pretreatment with IL-10 (10 ng ml^–1) and/or treatment with LPS (100 ng ml^–1), cells were treated with BFA for 1 or 2 h. Arrows in c–e indicate low and high modes of the TNF distributions. d The full TNF response to LPS requires intercellular communication. e Intercellular feedback through secreted TNF is necessary for the full response. Source data are provided as a Source Data file.

Fig. 2. Cell density modulates the heterogeneity of macrophage activation.

a Reporter protein fluorescence was measured by flow cytometry for the indicated cell densities, time points, and ligand treatments (IL-10 and sTNFR treatments as in Fig. 1). Percentages of highly activated cells were determined using a threshold (dotted vertical line) at the nadir between the two modes (arrows) of mCherry distributions. b Reporter trajectories for n = 30 cells at high density after treatment with sTNFR and LPS. Cells are ordered by cumulative mCherry expression and color-coded by fluorescence magnitude within heat maps. c Single-cell trajectories for total EGFP-RelA expression. The mean is in bold. d–e Relationship between initial and cumulative total EGFP-RelA (R² = 0.61, one-tailed permutation test p = 2 × 10⁻⁷) and between cumulative nuclear EGFP-RelA and cumulative mCherry (R² = 0.59, one-tailed permutation test p = 3 × 10⁻⁷). Dotted lines are linear fits, and axes are linearly scaled. In c–e, color-coding denotes rank-ordered cumulative mCherry expression. f Single-cell mCherry trajectories, with the mean in bold. Values are in a.u. specific to each panel. g Effect of culture density-associated conditions on macrophage activation heterogeneity. Fluorescence units are comparable within each reporter protein and passaging density. Dotted vertical lines distinguish low and high activation. h Revised conceptual model for macrophage activation with differently activated cell density-dependent subpopulations. Source data are provided as a Source Data file.

Fig. 3. A dynamical population model explains intracellular and extracellular signaling and regulation.

a The diagram summarizes variables, reactions, and mechanisms in the model of macrophage activation. Symbols: bold non-italicized text for variables; horizontal bars for compartment boundaries; n for nuclear and c for cytoplasmic; asterisks for activated receptors or kinases; delta symbol and circles for perturbation-specific effects; diagonal arrows for a perturbation’s effect; and graphs for time-dependent processes. b Generation of comparable distributions of initial values (data from Supplementary Fig. 2r). The observed distributions in (i) are in non-comparable units. To initiate simulations with NF-κB distributions that vary between high and low cell density and that match experimental observations for EGFP-RelA, we impute from an observed high density distribution (black) a low density distribution (teal) that matches the observed low density distribution (blue). The transform shifts the distribution (ii) and adjusts the proportion of cells in high vs. low states using a Gaussian mixture model of two populations fit to the high density distribution (iii), such that the combined transform (iv) generates an imputed distribution matching the observed low-density distribution (blue) with units relatable to high density (black). c–d Simulated TNF distributions from the calibrated model match experimental trends for cell density and ligand conditions in Fig. 1d, e and BFA conditions in Fig. 1c. All cells have the topology in a and are heterogeneous in initial value and transcription of Rela RNA and initial value of cytoplasmic NF-κB–IκB, derived from b. e–h Comparison of simulated and experimental outcomes for n = 30 cells at high density with sTNFR and LPS treatment. e–f Cumulative total EGFP-RelA (R² = 0.61, one-tailed permutation test p = 3 × 10⁻⁷) and cumulative nuclear EGFP-RelA (R² = 0.60, one-tailed permutation test p = 4 × 10⁻⁷), calculated using time-integrated values. The dotted diagonal is the identity line, and axes are linearly scaled. Color-coding is the same as in Fig. 2c–e. g Coefficient of variation (CV) in mCherry and total EGFP-RelA expression over time. h Trajectories were grouped post hoc by high or low activation. Mean values are in bold. Simulations are shown with and without FBD.

Fig. 4. Mechanisms of TNF regulation differ in phenotypic and functional consequences.

a Predicted and measured distributions of reporter expression across cell densities and doses of LPS or PMA at 12 hps. Very low density is 1/8th of low density. Arrows denote modes of the distributions for high density without stimulus or with 100 ng ml⁻¹ of LPS or PMA. For simulations: low and very low density used the same initial values; mCherry distributions without stimulus were obtained without FBD and without intercellular feedback; and FBD was applied for LPS doses at and above 1 ng ml⁻¹. b Simulated outcomes after individually varying the effect magnitudes of mechanisms that regulate TNF, with perturbations numbered and depicted in the diagram. Outcomes were assessed using the 12 hps time-integrated amounts of total EGFP-RelA, mCherry, and secreted TNF (cumulative secreted flux) per cell at high density. Conditions corresponding to the zero on the x-axis indicate base case effect magnitudes (1× on the color scale). Abbreviations: stabilizing regulation (SR) and destabilizing regulation (DSR). c Comparison of activation for homogeneous (hypothetical) or heterogeneous (observed) initial distributions of transcription factor expression. For each readout, a value of 1× was set for the outcome given a heterogeneous population at low density with low initial values. Homogeneous initial values were set to the mean of corresponding heterogeneous distributions. In the right panel, the effect of the number of cells is incorporated into the total amount of secreted TNF. Arrows indicate comparisons from the main text, and key trends are noted on the right. Source data are provided as a Source Data file.

Fig. 5. Primary macrophages regulate TNF production through quorum licensing.

a Bone marrow-derived macrophages exhibit cell density-dependent bimodality in LPS-induced TNF expression. The first two conditions (columns) are in biological duplicate and the other three conditions are in biological triplicate. b Secreted factors were measured in cell culture supernatants. The minimum value on each y-axis is the observed lower limit of detection of the assay. A subset of the panel is included here, and the full panel is in Supplementary Fig. 5d–f. Bar graphs represent the mean of the biological replicates and S.E.M. Effects of cell density and treatment condition were assessed using two-factor ANOVAs and Tukey’s HSD tests (Supplementary Note 2). Source data are provided as a Source Data file.